HMGforum.com

I and Luigi are trying to get HMG4 in better shape. I don't know if we will succeed but we are trying hard and got some results.

I was thinking about some problems:

1) HMG3 compatibility. I got some. We can go better. Personally, I'm not interested, since it is difficult to achieve and anyway, when you will be going to port your hmg3 program to hmg4 you need to check each program line to see if the function is implemented... at that point, just port code to HMG4 !
When I find HMG3 code snippets (like this one: viewtopic.php?p=17653#p17653 or the agenda demo alerady ported, really soon the contacts demo) I try to compile them with HMG4 and, depending on the problem found, add the missing bits to hmg4 or change the source code to be hmg4 compliant.

But is there anyone seriously thinking that he can compile and succesfully run its hmg3 code with HMG4 without any change ?

2) HMG4 compatibility with itself
We are doing some changes, I something may change... something already did... for example DEFINE BROWSE was using Browse() class and now uses VirtualGrid... Browse() class may be deleted from repository but... IS ANYONE ALREADY USING IT ?
Should I delete it or keep it ?

3) Using Qt directly
IS anybody using QtObject to access underlying Qt object ? It is a very interesting way to use HMG: you can add missing functionalities accessing directly the Qt object.... but what happens when we switch the Qt object used for a HMG object ?
For example I may want to port GRID (internally uses QTableWidget) to use ABSTRACTGRID (internally uses QTableView) to avoid duplicate code... if nobody uses it yet, it can be done easily, if someone uses it with HMG functions is ok, if someone access the underlying Qt object... it breaks !

4) is anyone really interested in HMG4 or have some software in production using HMG4 ?
This is probably the most important question of all: are we doing this work just for ourselves ? If there is no interest from others, we may take a different route in class management, widget handling, etc... to produce a framework that "looks like" HMG3 but points to be really OOP oriented... call it a fork if you prefer, or "HMG4 extended" to mimic what happened to hmg3.

I ask because HMG4 started as a good project, it quickly got developed, but in the hurry not every decision was "perfect"...


I and Luigi are spending really big time on HMG4 because we believe it can be a good base for writing "business" software. But are we alone in this ?

Mauricio, do you have software in the works based on HMG4 ?


Francesco

Hola Francesco:
realmente no soy muy bueno para escribir envolturas y me lleva mucho tiempo leer y comprender todos sus trabajos que hacen para HMG4.
Utilizo HMG3.x para mis trabajos, realmente es bueno y libre
hice algunas pruebas en HMG4, me quedo muy conforme con el desarrollo y puede dar mas, tiene muchas caracteristicas gracias a QT, se hacen muchas cosas buenas, realmente es un gran proyecto lo que estan encarando ustedes mis felicitaciones por ello.
solo se que el beneficio es hacia una gran comunidad que es HMG, no importando si es HMG3 o HMG4

Translate google
Hi Francesco:
I'm really not very good at writing wraps and it takes me a long time to read and understand all their work they do for HMG4.
HMG3.x I use for my work, it really is good and free
HMG4 did some testing, I am very satisfied with the development and can give more, has many features by QT are many good things, really is a great project that you are addressing my congratulations for that.
only that the benefit is to a great community that is HMG
regardless of whether or HMG4 HMG3

My opinion:
I see no benefits of QT-based HMG when using Windows. HMG4 programs are much slower then in HMG3, the dialogs don´t use the system language (you have to compile a version for every used language or write your own dialog) and the programs are very big.

HMG4 would be very interesting when Harbour/QT would run on mobile devices (Android, iOS) but i´m not sure if this is planned or available.

I totally agree with Daniel and Ricci.

Thank you!

Hi,

Yes. I am very much interested in HMG 4.

To share some of my thoughts:

- Basically I am a HMG user and not a HMG developer. Out of my curiosity and interest I have studied HMG source and changed something for some of my needs. Otherwise, I could not have known about the inner details of HMG.

- Regarding HMG4, IMHO, it is very nice since it makes free HMG from platform dependency.

- Regarding HMG3 compatibility, I think all of us would agree that, 100% compatibility will not be possible. However, the basic HMG functionalities are required to be compatible. The number of compatibility issues shall be reduced to the minimum possible. Some of the HMG controls are used very frequently (like button, textbox (with numeric/date/character inputmask), listbox, combobox, grids). Some controls/functionalities are not used frequently. The frequently used functions/controls shall have maximum compatibility.

- I am very much happy that yourself, Luigi and Mauricio Ventura Faria are doing a wonderful job in structuring and optimizing the code.

- I test the updated code frequently.

So, to conclude, let HMG3 be perfect for Windows environment and HMG4 shall be for all.

I'm totally agreed with Rathi

(Sorry, but except frequenty testing )

Best regards.

--

Esgici

Hi Ricci,
thank you for your message. It is really clear. I understand and I agree with your point of view.

As I said I don't earn an income any more from developing and selling software, it is more a hobby, I have just one big program still in production (text-mode) and some utilities for my own use.

So I understand that using a library "that works", trusted and proved by hundreds of programmers, with a working ide is a dream come true... it's not a problem if it is not OOP, if it uses clever preprocessor stuff or if it gives warnings at compile time... it works and works well ! And I'm happy with this. Really.

Ricci wrote: I see no benefits of QT-based HMG when using Windows.

I agree 100% ! Windows only program ? hmg3 is perfect for the job.

Ricci wrote: HMG4 programs are much slower then in HMG3,

Probably. Do you have some data ? I don't use hmg3 so I can't do comparisons. I also don't have "full featured" programs to do a compare...

the dialogs don´t use the system language (you have to compile a version for every used language or write your own dialog)

I will check but I believe there is a solution for this. Qt has a really powerfull translation subsystem and you can have one exe and load the translations from a file and witch between languages at runtime. There is the Qt Linguist application installed with Qt SDK to help create these translation files.

and the programs are very big.

In year 2011 4Mb against 2 Mb ? My phone has 16 GB of internal storage.... the other one has a 8GB sd card...

HMG4 would be very interesting when Harbour/QT would run on mobile devices (Android, iOS) but i´m not sure if this is planned or available.

the last versions of Qt are based on a new engine, it doesn't use any system widget but it creates its own simulating the look and feel of the original ones. In this way it can be "easily" ported to run on almost any s.o., it just needs to be able to draw on a screen (please have a look here: http://labs.qt.nokia.com/2008/12/02/wid ... -wolfenqt/)
So Qt android port is working, its not official but should be in little time, and I read also about a iOs version... but beware, you may find yourself with a "window" look on an iPad.... and infact I read about people saying that on android Qt doesn't use native widgets but its own, so it doesn't "feel" an android application... it will change of course (I hope )

Integrating harbour and Qt for android is a little mess... if it will ever work, probably hmg4 will need some changes because you can't have the same widgets on a full-hd screen or on a phone screen...

Ciao,
Francesco

rathinagiri wrote: To share some of my thoughts:

- Basically I am a HMG user and not a HMG developer. Out of my curiosity and interest I have studied HMG source and changed something for some of my needs. Otherwise, I could not have known about the inner details of HMG.

So you are now both a HMG3 user and developer.

- Regarding HMG4, IMHO, it is very nice since it makes free HMG from platform dependency.

Qt is a very complete library. It covers really different aspects in all supported platforms. Its layout and css systems are really spectacular... yes, css like for the web...

see http://thesmithfam.org/blog/2009/10/13/ ... tylesheet/ and http://thelins.se/learnqt/2009/05/qt-la ... he-basics/

- Regarding HMG3 compatibility, I think all of us would agree that, 100% compatibility will not be possible. However, the basic HMG functionalities are required to be compatible.

Ok, good. But I don't understand a point: if HMG3 users don't want to use hmg4 until it is complete and don't even want to take part at developing nor testing (I asked, one answer, no follow up)... who should develop HMG4 ?

The number of compatibility issues shall be reduced to the minimum possible. Some of the HMG controls are used very frequently (like button, textbox (with numeric/date/character inputmask), listbox, combobox, grids). Some controls/functionalities are not used frequently. The frequently used functions/controls shall have maximum compatibility.

Perfect ! minimum compatibility problems with most used widgets, so and so with the others.

SetProperty is a frequently used function, from mol post: In HMG4 should be, using a OOP syntax: I created a setProperty function like this: I understant why it is slower 3 macro in a line .... (we can also create one string and use one macro....)

But original setProperty function has up to 8 arguments, and big nested IFs... it is necessary to unroll the IFs and write code to test every branch...

But then we have _GetValue and _SetValue, a big array of _HMG_SYSDATA that uses a numeral index instead of "speaking" #defines, doMethod... getProperty with its nested IFs...

then in HMG3 there is @ syntax, DEFINE SYNTAX, and also alternate DEFINE SYNTAX, sometimes you can have ; sometimes not...

It can be done... but there is work for MONTHS !

So, to conclude, let HMG3 be perfect for Windows environment and HMG4 shall be for all.

It would be great!

Hi friends.
I'm very confused and very, very surprised that so few people are interested about this argument - 5 of 218 - and only
two of these might wish to develop HMG. On th other hand, I am very sad not to hear a comment by Roberto.

A lot of programmers have searched their "independence day" from MicroSoftWindows and now we want to take a strict link with GoogleAndroid?
But Android was Linux, now they have differences, because Google wants to be the next "big brother". On the other hand, I read many opinions about "in the next four years, they will be the same; another time".
I think Android is another way (for someone) to do business.

IMHO, I think the screen size about 8/10 inch can't be used to run business application form.
The programs are big? Yes, but we don't investigate how many things are linked using a simple hbmk.hbm file
The programs are slow? Perhaps, but what programs? Writed with HMG and HbQt? Writed with C++ and Qt? What are SysOp, processor, Ram, etc.
Dialog in language: yes can be a problem, but how many programs written with DBIII/Clipper language, can say "I'm in multilanguage" and OS dialog indepenedent?
And the last one: why HMG must run only on Windows system?

Anyway these are not "the question"; the question is: how many HMG3 users want to have an evolution and think HMG4 can be.

The silence that I hear is very eloquent and, given the efforts of many volunteers, makes me very sad to know - NOW - that a lot of people don't have interest.